Muni-Wi-Fi RevisitedFriday, January 11, 2008
Just when we thought the Muni-Wi-Fi hype was winding down, along comes a company with a "better idea" for Muni-Wi-Fi. I was (still am) convinced that Muni-Wi-Fi's time had passed or, rather, that there never was a time for it. The business models are questionable at best, maintenance issues could be very expensive as Wi-Fi grows in popularity and toward the end of last year there seemed to be industry-wide acceptance that Muni-Wi-Fi was dead on arrival. Most cities that have not deployed it have cancelled their plans to do so and a random check with the cities that have it up and running indicate a dearth of customers. EarthLink, AT&T and even Google had cooled off about it and San Francisco's system was doomed to failure before the first access point was installed.
Now a company called Meraki has offered to install a citywide system in San Francisco for free and to provide free access to any and all who want to use it, not at a reduced data speed of around 386 Kbps as others have promised, but at 1 Mbps.
Meraki has deployed its system in a test area within San Francisco and announced that it intends to blanket the city (roughly 50 square miles) with its system with no funding from the city government and no investment from others. It will fund the project itself with help from its venture partners. After I read about this endeavor, I contacted Sanjit Biswas, Meraki's co-founder and CEO, in order to better understand what the company intends to do and its business model.
What I learned is very interesting. First, access points won't be attached to lamp posts around the city, Wi-Fi repeaters will be placed in homes and businesses. Each repeater provides access to nearby repeaters and relays the signal on to the next repeater. At this point, Meraki is attaching about every third or fourth repeater to the Internet via its own back-end server system using DSL connections, and the equipment is furnished free to anyone requesting it (although on the Website the indoor repeater sells for $49 and an outdoor version sells for $99).
After reviewing the technology, which relies heavily on you and your neighbors to put up these repeaters but not to furnish any backhaul, I asked about the business model for San Francisco and was startled when Sanjit told me there is no business model. San Francisco will be a test bed for the systems, which are being marketed around the world. However, from the Meraki Website, it is obvious that there are several different business models including selling the hardware and back-end servers to others to deploy, embedding advertisements in the system and enabling services such as scrolling news and stock market results.
Funding has come from Google and Sequoia Partners, and the company just closed a round of $20 million. Meraki estimates that it will need to deploy 10,000 to 15,000 devices, meaning it will, essentially, have to go door-to-door to find volunteers to install the repeaters, but it feels it is up to the task and can make this work.
Meraki also claims to have a "magic sauce" to mitigate interference from other access points, and Sanjit agreed that from many places in San Francisco one can "see" 40 or more access points already in use. Even so, he is confident that Meraki can minimize interference while providing 1 Mbps service. The company has been running a trial in a much smaller area of San Francisco and has given away about 500 repeaters so far.
When I asked about customer support for the system, I was told that Meraki would not be providing voice technical assistance but would maintain a Website for frequently asked questions and some customer support. We will have to see about this one―even when people get something for free, they expect a certain level of support, and if they cannot get to the web how are they going to access what support is offered?
I am having a problem with the claim that this system can provide 1 Mbps service to everyone. Sanjit said that right now Meraki is using DSL connections about every third or fourth repeater, usually placed in businesses. Typical DSL speeds in San Francisco (according to the AT&T Website) run from 768 Kbps to more than 6 Mbps with pricing that runs from $15 per month to $35 per month. If Meraki does provide 1 Mbps per customer, it has to be using 3 or 6 Mbps DSL connections. Even with a favorable price from AT&T, 15,000 access points, with a connection to the Internet for every fourth repeater, means a total of at least 3,750 DSL lines. At $20 per month (a discount), the cost for DSL services alone is about $75,000 per month, and this does not include the cost of giving away 15,000 repeaters.
Another issue I have with this concept, which has been tried in Europe with some success and, according to Meraki, also in Brazil, is that you are at the mercy of your neighbors. If there is tension between neighbors and one of them decides to unplug the repeater, the result is a service outage. Also, when the power goes out, as it does during storms, every repeater in the affected area will stop working and the system will not be available.
While Meraki claims to have a proven way to mitigate interference, we are talking about the 2.4-GHz band, which supports only eleven channels for Wi-Fi and only three of these channels are non-overlapping. If only channels 1, 6 and 11 were used in a given area, they will not overlap with each other. However, channels 2, 3, 4 and 5 overlap with channels 1 and 6, and channels 7, 8, 9, and 10 overlap with channels 6 and 11.
When you have a large number of access points in a given area, all covering basically the same radius of 300 feet, there will be interference. This interference can be controlled fairly well on channels 1, 6 and 11, which are distinct channels, but when other access points are put into service on the other channels, there are many opportunities for interference. The average person who deploys one of these repeaters won't be aware of this potential interference, and while Meraki claims it can handle a lot of these issues with its "smart system back-end," there is a physical limit to what can be accomplished as interference continues to grow.
Most cable modem customers do not understand why their data rate appears to slow down in the afternoon and evening―it is because they are using shared bandwidth and more people are using it. Customers of the Meraki system will also be using shared bandwidth and will have varying results depending on the time of day, the number of customers sharing the repeaters between DSL drops and increased interference from new access points being installed. Today, most cable and DSL providers are offering a modem and router that includes both wired and wireless access, and in the tests we have run, it is not unusual to drive down a street and find an active access point at almost every house. When looking at apartments, it gets even worse. Further, there are competing technologies in use on the band: Bluetooth, cordless phones, wireless USB, Zigbee and, of course, microwave ovens.
Yes, I know the service is free, but when any service is installed there is an expectation that it will work and work as advertised. I wish both the city of San Francisco and Meraki good luck. If I lived in San Francisco, I would keep my DSL or cable service up and running.
This is a huge investment for a "test bed" approach to Muni-Wi-Fi, but I guess the venture funds are aware of this and believe it will provide a payback over time. While I am skeptical about this approach, it won't cost the city or its citizens any out-of-pocket money and the company's investors can easily afford this experiment, so no harm done here. Except, that is, if this system sets a level of expectations and does not deliver or does not work consistently, those who are getting free service will be reminded, once again, that there is no such thing as a free lunch.
Wi-Fi was invented to cover the last 300 feet, usually inside a building; it was not designed to be turned into a wide-area system. My Wi-Fi at home and at the office works great, and since it is tightly integrated with T-Mobile's wide-area network, I have the best of both worlds, including automatic Wi-Fi at home and anywhere T-Mobile has an access point, and where there is no Wi-Fi coverage, I have T-Mobile's wide-area network for my BlackBerry. Yes, it costs money, but I get good, reliable service, just as I do when I use my EV-DO-equipped notebook when on the road. San Francisco may be getting "free" Internet access, but I would rather pay for a service I can count on 24/7, 365 days a year.